Results
Sites used
Site A (Plates 2 & 3) was located at the Styx Mill Conservation Reserve (NZ Map Grid M36 77 921 E, 49 290 N), Site B (Plates 4 & 5) was located some 50m above the lower Styx Road bridge (NZ Map Grid M36 78 865 E, 48 898 N), and Site C (NZ Map Grid M36 79 277 E, 49 122 N; Plates 6 & 7) was located approximately 300m upstream of the railway bridge. Full site descriptions are provided in Appendix A.
Table I. Data collected from three sites on seven sampling occasions.
Plate 2. Looking upstream from Site A.
Plate 3. View from true left bank at Site A.
Plate 4. Looking upstream from Site B.
Plate 5. View from true left bank at Site B.
Plate 6. Looking upstream from Site C.
Plate 7. View from true left bank at Site C.
Site monitoring data
While mean flows, turbidities, and suspended solid concentrations all tended to increase downstream (Table I), the inherent variability was too great for significant differences to be detected between the sites (ANOVA, p>0.05). Mean velocities were significantly lower at Site C, and the water was significantly deeper, than at the upstream sites (ANOVA, p<0.05; Duncan Test).
Suspended Solids
The relationship between streamflow (m3s-1) and suspended solids (mgL-1) for each site is shown in figure 3. While the relationships generated at each site describe the data well, inherent variability in the data between sampling dates meant that 95% confidence intervals for the relationships overlapped and consequently are not statistically different from each other.
Site A..... SS (mgL-1) = 2.8e1.8939x (R2=0.61; ANOVA, p<0.05)
Site B..... SS (mgL-1) = 3.795e1.3454x (R2=0.70; ANOVA, p<0.05)
Site C..... SS (mgL-1) = 1.68e1.66x (R2=0.92; ANOVA, p<0.01)
where x = flow (m3s-1)
Figure 3. Suspended Solid Concentrations with flow at the three sites.
Figure 4. Organic content of suspended solids with flow at the three sites.
Equivalent relationships between flow and mean velocity, and flow and turbidity, are included in Appendix B, but were not used in further analysis. No relationship was found to occur between river flow and the organic fraction of the suspended solids (Figure 4; ANOVA, p>0.05).
Specific Stream-Segment Relationships for Velocity and Depth
Relationships between flow and velocity, and flow and depth, for each 1m wide segment of each site are summarised in Appendix B.
River flow data
One thousand, seven hundred and thirty daily mean-flow recordings at Radcliffe Road were classified into 0.2 m3s-1 increments. The number of days in each flow-increment was divided by the period covered (years) to obtain the annual occurrence of each. This was then plotted as a frequency-flow hydrograph (Figure 5).
Figure 5. Annual frequency-flow hydrograph for the Styx River at Radcliffe Road.
Sediment Deposition Rates
Sediment accumulation in the sediment traps was significantly higher at the fastest velocity used (0.8ms-1) than at slower velocities (Figure 6; ANOVA, p<0.01). However, this treatment was the only one that did not have baffles to manipulate water velocities. Significant amounts of weed were found to accumulate during the 24 hour sampling period on the baffles and when the weed clumps were removed, sediment was dislodged from the weed matrix and visible plumes of sediment were observed. Clearly the weed may have had a significant filtering action on the water passing over all the treatments except the 0.8ms-1 treatment, and probably the 0.0 ms-1 treatment also (because of lack of flow). Consequently, the 0.8ms-1 was used as a pseudo control for the effect of baffles and weed filtration on the other treatments. This was achieved by forecasting the trend observed in the other treatments to predict accumulation in a hypothetical weed-affected 0.8ms-1 treatment (c. 22 g.day-1). The ratio between the measured, unaffected treatment to the hypothetical, weed-affected treatment was then used to calculate the increased sedimentation that would have occurred in the other treatments if baffles were not used (since natural redds do not have baffles). This correction increased sedimentation rates by a factor of 3.86 in the other treatments (Fig. 6). Using the corrected data, sedimentation rates were highest at velocities below approximately 0.4 ms-1, and levelled off between 0.4 ms-1 and 0.8ms-1. Although velocities above 0.8ms-1 were not measured, observations of scouring during the flood suggested that velocities above 1.4ms-1 caused net removal of material in traps. This would mean that sedimentation rates reduce to zero between 0.8ms-1 and about 1.4ms-1. Consequently, a polynomial trendline was used to estimate sedimentation rates within this range. For velocities above 1.4ms-1, a net removal rate of 45 g.day-1 was used to model sediment removal from hypothetical traps (0.04m2).
Figure 6. Amount of sediment deposited in 24 hours versus velocity, and the correction applied to account for weed effects.(Area=0.04m2; SS=9.5mgL-1; Error Bars = ±1 SE)
When the sediment is expressed as a percentage of the total suspended load passing over the bed area, the deposition rates are highest at low velocities and quickly decline as velocities increase (Figure 7).
Figure 7. Percentage of the suspended load that is deposited, versus velocity.
Significant differences in organic content of the deposited solids were found between the two slowest velocities (0 and 0.2 ms-1) and the two fastest velocities (0.6 & 0.8ms-1). The amount of organic content of sediment deposited at 0.4ms-1 was only significantly different than sediment accumulated at 0ms-1 (ANOVA, p<0.01; Duncan Test). Generally, less organic material is deposited at higher water velocities (Figure 8).
Figure 8. Organic content of the deposited solids, versus velocity.(Error bars ±1 SE)
Sediment accumulation modelling
Modelling Flows
Table II. Flows used at each site in a 76 day incubation period (days rounded to the nearest day).
Table II shows the range of flows at each site that were used in the model, and the frequency (days) of each flow, during the incubation period.
Modelling Deposition Rates
The mathematical function used to calculate daily deposition in each segment of the stream bed, at each site, followed the following form:
Daily deposition (g/m2/day) = SS / SSEXPT * Deposition Rate (g.m-2day-1) * Depth/DepthEXPT
The first term in the equation is the ratio of suspended solid concentrations to the concentration that was present during the measurement of deposition rates. This assumes that the relationship between deposition rate and velocity is directly proportional to the amount of sediment in the water. Similarly, the last term assumes that sedimentation increases in proportion to water depth.
The deposition rate is the total amount of sediment that was accumulated in the traps, in a day. The equation used to determine deposition rate as a function of water velocity (Fig. 6) was:
Mass deposited =
| If v > 1.4ms-1, | then y = | -45 g.m-2day-1 | |
| 0.04 | |||
| if v < 1.4ms-1, | then y = | 750v5-2808v4+3705v3-2097v2+440v+88 g.m-2day-1 | |
| 0.04 |
v = velocity (ms-1) in each section of the river bed.
To obtain estimates for organic material, the total sediment was multiplied by an equation derived from Figure 8 that calculates the proportion of sediment which is organic.
Organic = Total Sediment * 12.44* e-1.972*v
100
Modelling sedimentation during redd incubation
A spreadsheet was developed to calculate the daily accumulation of sediment, and organic sediment, in each stream segment at each flow increment. The spreadsheet was also used to randomly assign days to each of the expected flow events (to determine when they would occur during a hypothetical incubation period). Because sediment accumulation is likely to be reduced as superficial gravel spaces become clogged (armouring), and no data to determine the rate of this process had been gathered, a range of armouring rates as a function of time, were modelled. Rates used in the model varied from no armouring effects (y=e0) to a rapid decline in deposition rates over the first few days (y=e-0.14).
While the actual amount of sediment accumulating at each site could not be predicted without guessing at armouring effects, the relative sedimentation between sites was the same in all scenarios modelled. Regardless of how quickly armouring reduced sedimentation rates, sediment accumulation at Site A was 76% relative to Site B, and 23% compared to Site C. Similarly, accumulation at Site B was 30% of that at Site C.
If sedimentation rates did not change with time, the model predicted that approximately 268 kg.m-2 would accumulate at Site A, 352 kg.m-2 at Site B, and 1193 kg.m-2 at Site C. However, when the effects of armouring were included these estimates dropped to much lower levels (Figure 9).
Figure 9. Predicted final sediment loading at three sites depending on how quickly armouring effects reduce deposition rates. Values below the dashed line probably represent situations where redds would be viable.
Figure 10. Predicted organic sediment accumulation, at three sites, depending on how quickly armouring affects deposition rates.
Organic sediment deposition was predicted to be similar at Sites A and B, but much higher at Site C (Figure 10).
A best-guess estimate using armouring rates
To obtain an estimate of total sediment accumulation during the incubation period, the effect of armouring on deposition rates e-0.1*time(days) was used. Based on this rate the amount of sediment accumulated at each site is shown in Table III.
Table III. Amount of sediment predicted to accumulate during incubation, where armouring reduces sediment deposition during incubation at the rate e-0.1*time(days).
Site A
Site B
Site C
Total (kg.m-2)
38
50
171
Organic (kg.m-2)
1.2
1.3
9.8
Figure 11. Hypothetical decline in sediment deposition rate through time (due to armouring) if decline = e-0.1*time (days)